What’s a seed in a random number generator?

In reality pseudo­random numbers aren't random at all. They are computed using a fixed determi­nistic algorithm.

The seed is a starting point for a sequence of pseudorandom numbers. If you start from the same seed, you get the very same sequence. This can be quite useful for debugging.

If you want a different sequence of numbers each time, you can use the current time as a seed.

Example

This generator produces a sequence of 97 different numbers, then it starts over again. The seed decides at what number the sequence will start.

// New returns a pseudorandom number generator Rand with a given seed.
// Every time you call Rand, you get a new "random" number.
func New(seed int) (Rand func() int) {
    current := seed
    return func() int {
        next := (17 * current) % 97
        current = next
        return next
    }
}

func main() {
    rand1 := New(1)
    fmt.Println(rand1(), rand1(), rand1())

    rand2 := New(2)
    fmt.Println(rand2(), rand2(), rand2())
}

17 95 63
34 93 29

The random number generators you’ll find in most programming languages work just like this, but of course they use a smarter function. Ideally, you want a long sequence with good random properties computed by a function which uses only cheap arithmetic operations. For example, you would typically want to avoid the % modulus operator.

Share: